StatCon Workshops 2023
31 August - 2 September, 2023
Workshop Extreme Value Statistics
John H.J. Einmahl
We begin with a detailed introduction to univariate and multivariate extreme value statistics. Since we assume basic knowledge of univariate extreme value theory, we put emphasis on the multivariate case, which will be for notational convenience the bivariate case. We will discuss the extreme value index and tail heaviness as well as the estimation of extreme quantiles for univariate data and tail dependence and tail copula estimation for bivariate data. We will sketch some of the many applications the theory has.
After the introduction we will discuss three recent papers:
- 1Limits to human life span through extreme value theory (with Jesson Einmahl and Laurens de Haan). This applied paper answers the question if “the probability distribution of the human life span has a finite endpoint or not and, if so, whether this upper limit changes over time”.
- Extreme value inference for heterogeneous power law data (with Yi He). This paper considers univariate data from possibly different distributions, with an application to assess the tail heaviness of earthquake energies. We also show the good finite-sample behavior of our limit theorems through simulations.
- Tail copula estimation for heteroscedastic extremes (with Chen Zhou). The paper considers tail dependence estimation, allowing the marginal distributions to be different according to a so-called scedasis function. Again, the asymptotic theory is supported by simulations.
Workshop An introduction to interval censoring
Ingrid van Keilegom
Workshop The k-sample problem: a brief review and some new issues
University of Seville
The k-sample problem consists in testing for the equality of k distributions based on independent samples from each population. This is a well-studied problem which has generated a vast literature. Most existing techniques assume a ﬁxed number of populations, with arbitrary but speciﬁed dimension, and that the sample sizes increase (the classical setting). This problem still generates many articles which try to adapt the existing techniques and/or propose novel methods to new settings such as high-dimensional data, where the dimension is larger than the sample sizes, and large k, where the number of populations is larger than the sample sizes, among others. This course brieﬂy reviews procedures designed for the classical setting and discusses new tests for the case of large k.
Workshop The missing data problem(s) and solutions
University of Belgrade
The problem of missing data is one of the most common in all research that relies on data. Therefore, it is very important to adequately solve it.
The course will provide an introduction to different types of missingness, as well as the most common methods for solving this problem. The strengths and weaknesses of each method will be discussed, depending on the types of missingness. Everything will be illustrated with practical examples in the R programming language. Therefore, familiarity with R is recommended.
Workshop User-friendly biplots in R with biplotEZ
Centre for Multi-Dimensional Data Visualisation
Biplots are valuable visualisation tools in exploratory data analysis. In its simplest form, biplots are regarded as generalised scatterplots for more than two variables. The availability of software limits biplot application to expert users. Providing an EZier to use package for practitioners wanting to visualise their data, encouraged the development of a user-friendly R package. This is currently the flagship project of the Centre for Multi-Dimensional Data Visualisation (MuViSU) at Stellenbosch University. In this workshop you will be introduced to the main aspects of biplot methodology and receive access to the newly developed functions of the biplotEZ R package with applications on real data in various contexts.
Workshop Directional data analysis
Karlsruher Institut für Technologie (KIT)
We introduce directional data as multivariate data conditioned to the assumption of unity of length of the data vector. We start the workshop by showing applications of such modelled data from various fields such as geology, astronomy, or paleo magnetics. After introducing spherical versions of summary statistics such as measures of location, concentration, and dispersion, we focus on basic concepts like characteristic functions, limit theorems, and models of directional distribution theory. Finally, we introduce concepts of point estimation and statistical tests with an emphasis on uniformity testing.
Workshop Life in Academia
National and Kapodistrian University of Athens
This will be a panel discussion on life in academia.
Workshop Data Science Collaborations
University of the Western Cape
This will be a panel discussion on Data Science Collaborations.